Does Copying Congress's Stock Trades Work? We Tested 4,967 Buys
· Updated
We took 4,967 stock purchases disclosed by members of Congress between June 2024 and September 2026 and "bought" each one the day after it became public. Six months later, the typical copied trade had trailed the S&P 500 by 3.5 percentage points. Only 42% beat the index.
Members of Congress disclosed 4,967 stock purchases between June 2024 and September 2026. Picture them as 100 dots.
We copied each one the day after it went public. Six months later, only 42 in 100 had beaten the S&P 500.
Pull the winners aside and the losers are the bigger pile.
The typical copied trade trailed the index by 3.5 percentage points.
Spot on.Close, a little low.Close, a little high.Way off, too low.Way off, too high. It was −3.6 pts.
It trailed by 3.6 points, almost exactly the 3.5 points a copier lost buying after the filing. There was no edge for the delay to erase.
The disclosure delay is not why. Buying on the day the member actually traded, weeks before anyone could see it, gave almost the same result: the typical buy trailed by 3.6 points. The trades themselves don't beat the market. Below: the full results, the members and trade types we split out, Nancy Pelosi's record, and what to do with a disclosure instead.
A member of Congress buys a stock. Nobody else can see the trade yet.
The filing lands weeks later, a median of 29 days. Only then can a copier buy.
Hold both for six months and compare each with the S&P 500 over the same stretch.
The member's own entry trailed by 3.6 points. The copier's trailed by 3.5. There was no edge for the delay to erase.
The short version
- Copying lost to the index. Entered the day after filing, the median buy trailed the S&P 500 at every horizon we tested: 1.9 points at 90 days, 3.5 at 180 days, 8.1 at a year.
- The 45-day delay barely matters. Entering on the member's own trade date did no better. There was no edge for the delay to erase.
Day one: the copied buy and the index start level.
After 90 days the median copied buy trailed by 1.9 points.
After 180 days, by 3.5 points.
After a year, by 8.1 points. Copying lost to the index at every horizon we tested.
- Better than random, not better than the index. Congress's buys lag the index by less than random stocks bought on the same days, mostly because members favor the mega-caps that led the market. On average, their picks did no better than random ones.
- No filter rescued it. Big trades, spouse trades, fast filers and Senate trades all landed within a few points of the index, and none was consistent across horizons.
- Pelosi is a coin flip too. Of 17 spouse purchases we could price at six months, 6 beat the S&P 500 after the filing date.
How we tested it
We used Yield Theory's congressional-trades dataset, which is parsed from House Clerk and Senate eFD filings and keeps the transaction date, filing date, owner and amount range on every row.
- Trades: every disclosed purchase with a stock ticker, transacted from June 6, 2024 to September 22, 2026 and filed by October 2, 2026. We counted a member buying the same ticker on the same day once. That left 4,967 purchases by 82 members across 1,103 tickers.
- Prices: daily closes adjusted for splits and dividends, so both the stock and the index include dividends.
- Two entry points: the close on the transaction date, which is what the member got, and the close on the first trading day after the filing date, which is the earliest a copier could act.
- Benchmark: SPY, an S&P 500 index fund, bought on the same day as each trade.
- Horizons: 30, 90, 180 and 365 calendar days. A trade only counts at a horizon once that much time has passed, so the sample shrinks for longer horizons.
- Member trades (Day 0) to Trade-date entry (What the member got): Same day
- Member trades (Day 0) to Filing-date entry (Earliest a copier could act): Median 29 days
- Trade-date entry (What the member got) to Compare with SPY (30, 90, 180, 365 days)
- Filing-date entry (Earliest a copier could act) to Compare with SPY (30, 90, 180, 365 days)
"Excess return" below means the stock's return minus SPY's over the same window. If a stock rose 10% while SPY rose 13%, the excess return is −3 points.
Result 1: copied trades trailed the S&P 500
| Horizon | Trades priced | Median return | Median excess vs S&P 500 | Average excess | Share that beat the S&P 500 |
|---|---|---|---|---|---|
| 30 days | 4,760 | +1.5% | −0.6 pts | +0.4 pts | 46% |
| 90 days | 4,203 | +3.8% | −1.9 pts | +0.2 pts | 45% |
| 180 days | 3,742 | +6.9% | −3.5 pts | −0.5 pts | 42% |
| 365 days | 2,527 | +14.3% | −8.1 pts | +2.1 pts | 42% |
Entry on the first trading day after the filing date.
The median and average tell different stories. The median copied trade lost to the index at every horizon, and it lost by more the longer it was held. The average sits near zero because a handful of big winners offset the many trades that lagged.
That shape matters for anyone copying with real money. If you copied every trade, the winners roughly paid for the losers and you'd have ended up near the index, holding hundreds of positions with more risk and more taxable trades. If you copied a few trades, the odds were that most of yours lagged.
Result 2: the delay wasn't the problem
The most common objection to copying Congress is that filings arrive too late. In this sample the median purchase was disclosed 29 days after the trade, and 26% took longer than 45 days. So we reran everything as if you could have bought alongside the member on the transaction date.
| Horizon | Median excess, trade date | Median excess, filing date | Beat rate, trade date | Beat rate, filing date |
|---|---|---|---|---|
| 30 days | −0.8 pts | −0.6 pts | 46% | 46% |
| 90 days | −2.0 pts | −1.9 pts | 44% | 45% |
| 180 days | −3.6 pts | −3.5 pts | 43% | 42% |
| 365 days | −9.3 pts | −8.1 pts | 40% | 42% |
The two columns are nearly identical. Buying with zero delay would not have helped, because the trades didn't beat the market in the first place. A faster tracker or a real-time alert service doesn't fix that.
Result 3: slightly better than random picks
Trailing the S&P 500 isn't unusual for single stocks. Over this period a few mega-caps drove much of the index's 48% total return, so most individual stocks lagged it. The fairer question is whether members picked better than chance.
To check, we swapped each purchase for a randomly chosen stock from the same pool of 1,103 tickers, bought on the same day, and repeated that 200 times.
| Horizon | Congress: median excess | Random: median excess | Congress: beat rate | Random: beat rate | Congress: average excess | Random: average excess |
|---|---|---|---|---|---|---|
| 90 days | −1.9 pts | −2.9 pts | 45% | 42% | +0.2 pts | −0.3 pts |
| 180 days | −3.5 pts | −5.0 pts | 42% | 41% | −0.5 pts | 0.0 pts |
| 365 days | −8.1 pts | −10.2 pts | 42% | 39% | +2.1 pts | +4.3 pts |
Entry on the first trading day after the filing date for both.
At the median, Congress's buys did lag by 1 to 2 points less than random stocks. Most of that comes from what they buy: NVIDIA, Microsoft, Amazon, Apple and Meta were the five most-purchased tickers, the same names that carried the index. On average, though, random picks did as well or better. Members tilt toward big, popular companies, and that's about all the data shows.
Result 4: no filter fixed it
Tracker sites often suggest following only certain trades. We tested the common filters, entered on the day after filing:
Tap the line you think is the tell.
Spotted it.Not that line. Big trades were the only filter ahead at 90 days, on about 200 trades. Check how it held up at 180 days before you trust it.
| Filter | Trades at 90 days | Median excess, 90 days | Beat rate, 90 days | Median excess, 180 days | Beat rate, 180 days |
|---|---|---|---|---|---|
| All purchases | 4,203 | −1.9 pts | 45% | −3.5 pts | 42% |
| Senate only | 263 | −0.3 pts | 48% | −0.4 pts | 49% |
| House only | 3,940 | −1.9 pts | 45% | −3.9 pts | 42% |
| Member's own account | 2,469 | −1.7 pts | 46% | −4.8 pts | 41% |
| Spouse's account | 1,115 | −2.7 pts | 43% | −2.6 pts | 45% |
| $50,001 or more | 219 | +1.9 pts | 57% | −0.6 pts | 48% |
| $1,001–$15,000 | 3,507 | −2.3 pts | 44% | −3.7 pts | 42% |
| Filed within 14 days | 766 | −1.4 pts | 47% | −3.0 pts | 43% |
| Filed after 30 days | 1,747 | −3.1 pts | 42% | −3.3 pts | 43% |
The large-trade filter is the only one that beat the index, at 90 days, and its edge disappeared by 180 days on a sample of about 200 trades. Test enough filters and one will look good by chance. A real edge would hold up across horizons.
Weighting each member equally gives the same answer. Among the 48 members with at least five priced purchases, the typical member's average copied trade beat the S&P 500 by 0.1 points over 90 days. Half the members beat it and half didn't.
Individual members: mostly below the index
Prolific traders dominate any copy strategy, so we looked at the members with the most purchases. Six-month results, entered the day after filing:
| Member | Purchases | Priced at 180 days | Median excess | Share that beat the S&P 500 |
|---|---|---|---|---|
| 905 | 583 | −7.9 pts | 38% | |
| 700 | 680 | −1.9 pts | 46% | |
| 238 | 236 | −0.6 pts | 49% | |
| 229 | 228 | −8.7 pts | 37% | |
| 175 | 135 | −2.2 pts | 44% | |
| 139 | 133 | +8.5 pts | 68% |
Tim Moore is the standout, with 68% of 133 copied buys beating the index. Picked out after the fact from dozens of members, one strong record is expected even with no skill involved. His purchases also cluster in time, so they aren't 133 independent tests. Treat it as a reason to look at his filings, not a strategy.
What about Nancy Pelosi?
"Pelosi tracker" is the most-searched version of this idea. Every Pelosi purchase in our window was reported under her spouse, Paul Pelosi, and they're large: 22 purchases, mostly in the $250,001 to $5,000,000 bands. Many are call options or option exercises rather than share purchases, so the figures below show how the underlying stocks did after each filing, not what the household earned. Our Pelosi portfolio breakdown separates the two.
| Horizon | Purchases priced | Median excess, filing date | Beat the S&P 500 |
|---|---|---|---|
| 90 days | 19 | +1.4 pts | 11 of 19 |
| 180 days | 17 | −3.0 pts | 6 of 17 |
| 365 days | 8 | +15.2 pts | 5 of 8 |
The one-year figure looks impressive, but it comes from 8 purchases, several of them the same stocks bought the same day (Alphabet, Amazon and NVIDIA in January 2025). At six months, the larger sample, most copied buys lagged. Twenty-two trades in a few mega-cap names can't separate skill from a good run in big tech. The Pelosi trade page lists every filing with its source document.
The funds that automate this have the same problem. Our NANC and KRUZ breakdown shows each congressional-trading ETF beating the S&P 500 over one period and trailing it over another, while charging about 0.72% a year.
Why the trades don't beat the market
The data can't show motive, but it does fit a simple explanation.
- Most trades are routine. 84% of purchases were in the smallest band, $1,001 to $15,000. Many look like portfolio maintenance, reinvestment or adviser-run accounts, not conviction bets.
- Many accounts are managed by someone else. Spouse, joint and dependent accounts made up 39% of purchases, and members often say an adviser makes the trades.
- Ordinary stock picking is hard. Most single stocks trailed the index over this period. A member of Congress choosing well-known companies faces the same odds as anyone else.
None of this rules out a well-timed trade. It means the average disclosure carries no signal you can trade on.
Our read
Don't copy congressional trades. Across almost 5,000 purchases, copying gave you roughly the index on average, worse than the index in the typical case, and a lot more positions to manage. Simply buying the S&P 500 was the better strategy, and it's the bar any stock pick has to clear: over the past ten years, only 28% of today's S&P 500 companies cleared it.
- Winners roughly paid for the losers
- Near the index on average
- Hundreds of positions, more taxable trades
- Odds were most of yours lagged
- Median trailed by 3.5 pts at six months
- One position
- The bar any stock pick has to clear
A disclosure is still useful as a prompt. When a member buys a stock, you have a name, a date and a public record. What you don't have is a reason to own it. That takes research the filing can't give you: what the business earns, what it's worth and what has to go right.
Try it yourself
- Open the
recent purchases list and pick one trade.
- Note the transaction date, filing date and amount range. The
disclosure guide explains each field.
- Open the
stock page for the ticker and look at its revenue, margins and cash flow.
- Write one sentence on why the business could beat the S&P 500 from here. If you can't, the trade isn't telling you anything.
Questions readers ask
Is there a congress trading strategy that beats the market?
Not in this data. We tested chamber, owner, trade size and filing speed, and every filter landed within a few points of the index, with no edge that held across horizons. Earlier academic studies that found an edge mostly used data from before the 2012 STOCK Act.
Would a real-time alert app help?
No. Buying on the member's own transaction date, before anyone could see the trade, gave almost the same result as buying after the filing. Speed doesn't add returns the trades never had.
Are Pelosi's trades worth following?
Her disclosed purchases are large and concentrated in big tech, which did well over this period. At six months, though, only 6 of 17 copied buys beat the S&P 500. The sample is too small to show skill.
Where does the data come from?
Official House Clerk and Senate eFD disclosures, parsed into our congressional-trades dataset, which you can download as CSV or JSON. Each row links to its source filing.
Go deeper with member research
Copying a trade outsources the decision. These show how we make it ourselves:
Issue 05: NVIDIA Is Building the Credit Market for AI: NVIDIA was Congress's most-bought stock in this sample. This issue looks at how its customers pay for GPUs and where that financing could go wrong.
Intel's manufacturing option needs a cash-flow bridge: Paul Pelosi bought Intel twice in 2026. This case study shows what Intel's recovery evidence actually said on January 23, 2026.
The fundamental research course: 40 chapters, from reading a 10-K to writing your own research memo. The first chapter is free.
Limits of this data
- The window, June 2024 to October 2026, was a strong market led by a few mega-caps. Results could differ in a falling market or a broader rally.
- Amount ranges mean we can't weight trades by dollars, so every purchase counts equally.
- We tested purchases of stock tickers only. Options, bonds, funds without tickers and private assets are excluded, and sales are not netted against purchases.
- Prices are daily closes. A copier would pay a little more or less depending on when they traded, plus any trading costs and taxes.
- Members file amendments, and late filings can add trades to past periods. Figures reflect filings available on October 2, 2026.
Companies in this article
Keep reading
Members-only research
Members are reading this
- Issue 09: Is Nvidia Stock a Buy at 10 Times Earnings Without Huang?
At $155.46, Nvidia trades at 9.8 times next year's expected earnings, about half the S&P 500's multiple. What the price assumes, and our three calls.
- Issue 04: Citadel Bought Time at a 10% Discount
The buyer did not need a better AI forecast. Citadel was paid to provide liquidity when Situational Awareness could no longer finance the path to its own long-term thesis.
The thesis, the numbers behind it, and what would break it. Full access is $39 a month.
Read the full researchGet a full issue free
Enter your email and we'll send you a complete Yield Theory issue, then the opening of each new one. Unsubscribe anytime.